• About Me

    I received my B.Sc. degree in Computer Science at Amirkabir University, Iran, 2011. In my B.Sc. thesis, I worked on developing an API for Data Mining Query Language with SQL Server. I received my M.Sc. degree in Computer Science at Amirkabir University, Iran, 2013. In my M.Sc thesis, I worked on developing a framework for improving a transportation safety problem using decision tree rule induction and neural networks. I started my PhD study in Computer Science at University of California Irvine, 2014. In my PhD thesis, I've been working on developing distributed and large-scale solutions for spatio-temporal problems. First, I developed distributed optimization algorithms for network flow problem. Second, I developed deep learning models for general spatio-temporal problems, such as forecasting, missing data imputation, clustering and anomaly detection problems.

    Download CV

    Contact address: rasadi [at] uci [dot] edu or LinkedIn

    .
My Specialty

My Skills

Python

100%

Java

100%

c++

75%

AWS

75%
Education

Thesis:"Deep Learning Models for Spatio-Temporal Data Forecasting and Analysis" GPA:3.9/4.0

Amirkabir University, Iran, 2011-2013

  • Research Areas: Machine Learning in Transportation Problems
  • Thesis: A Decision Tree based Rule induction System for Transportation Safety Prediction.

Amirkabir University, Iran, 2007-2011

  • Research Areas: Search Engine Optimization and Data Mining
  • Thesis: A Data Mining QueryLanguage API for SQL Serve

Experiences, Publications and Projects

A rule-based Decision Support System in Intelligent Hazmat Transportation Systems

To improve the performance of accident severity prediction, we investigate rule extraction methods from decision trees and learning extracted rules with neural network. During this project I work with various techniques of decision trees for generating if-then rules, sparse rules with neural networks, and data mining approaches on accident data. The results is published in IEEE Transaction on ITS 2015, here

.

A Spatial-Temporal Decomposition Based Deep Neural Network for Time Series Forecasting

In this paper, we formulate spatio-temporal forecasting problem. We design a deep learning model which efficiently consider similarities in spatial data, various periodic patterns in temporal data, and is robust to missing data. The implementations are with Python-Keras. During this project, I work with various techniques for time series forecasting, time series decompositions, and convolutional-recurrent neural networks. The results is published in here

.

A convolutional recurrent autoencoder for spatio-temporal missing data imputation.

In this project, we design a convolutional-LSTM neural network for missing data imputation of spatio-temporal data. We illustrate that convolutional-LSTM architectures are stronger models to capture spatial and temporal patterns together. The model efficiently impute missing values from such datasets. During this project, I work with various baseline missing data imputation techniques, convolutional and recurrent layers, and latent feature representation of authoencoders. The results is published in International Conference on AI, 2019 here and "A convolutional recurrent autoencoder for spatio-temporal missing data imputation"; Reza Asadi, Amelia Regan; 21st International conference on artificial intelligence;

.

Distributed convex optimization solution for network flow problem

We developed distributed optimization algorithm for optimal network flow problem, by reducing the number of decision variables using the graph theory. The model has a tremendous improvement compared with classifical optimal network flow methods. The results is published in here

Also, the application of such solution on power grid networks is presented and accept in . "Cycle flow formulation of Optimal network flow problems and respective distributed solutions";Reza Asadi, Solmaz Kia; Journal of Automatica, Sinica

Work Experiences

Teaching Assistant

Deep Learning and Neural network, Winter 2019, professor Baldi

Artificial Intelligence, Fall 2018, Professor Kask

Advanced Data Structure, Spring 2018, professor Dillencourt

Artificial Intelligence, Winter 2018, Professor Lathrop